JET2000: aircraft observations of the African Easterly Jet
system C D Thorncroft, D J Parker
The JET2000 project: Aircraft observations of the African
easterly jet and African easterly waves.
Thorncroft, C. D.1, Parker, D. J.2, Burton, R.R.2, Diop, M.2,
Ayers, J. H.3, Barjat, H.4, Devereau, S., Diongue, A.2, Dumelow,
R.5, Kindred, D.R.4, Price, N.M.4, Saloum, M.6, Taylor, C.M.7,
Tompkins, A.M.8
1 Department of Earth and Atmospheric Science, University at
Albany, Albany, New York, USA
2 Institute for Atmospheric Science, School of the Environment,
University of Leeds, Leeds, UK
3 HATS (MRF), DERA, Boscombe Down, UK
4 The Met Office, DERA, Farnborough, UK
5 The Met Office, Bracknell, UK
6 The Niger Met Service, Niamey, Niger
7 Centre for Ecology and Hydrology, Wallingford, UK
8 ECMWF, Reading, UK
Submitted to Bulletin of the American Meteorological Society
Corresponding author: Chris Thorncroft
e-mail: [email protected]
Abstract
Scientific background and motivation for the JET2000 aircraft
observing campaign that took place in West Africa during the last
week of August 2000 is presented. The Met Research Flight C130
aircraft made two flights along the African easterly jet (AEJ)
between Sal, Cape Verde and Niamey, Niger and two ‘box’ flights
that twice crossed the AEJ at longitudes near Niamey. Dropsondes
were released at approximately 0.5-10o intervals. The two ‘box’
flights also included low-level flights that sampled north-south
variations in boundary layer properties in the baroclinic zone
beneath the AEJ.
Preliminary results and analysis of the JET2000 period including
some of the aircraft data are presented. The JET2000 campaign
occurred during a relatively dry period in the Niamey region and,
perhaps consistent with this, was also associated with less
coherent easterly wave activity compared to other periods in the
season. Meridional cross-sections of the African easterly jet (AEJ)
on the 28th and the 29th (after the passage of a mesoscale system)
are presented and discussed. Analysis of dropsonde data on the
28th, indicates contrasting convective characteristics north and
south of the AEJ with dry convection more dominant to the north and
moist convection more dominant to the south. The consequences of
this for the AEJ and the relationship with the boundary layer
observations are briefly discussed.
Preliminary NWP results indicate little sensitivity to the
inclusion of the dropsonde data on the AEJ-winds in ECMWF and Met
Office analyses. It is proposed that this may be due to a good
surface analysis and a realistic model response to this. Both
models poorly predict the AEJ in the 5-day forecast indicating the
need for more process studies in the region.
The JET2000 aircraft campaign in West Africa is introduced along
with preliminary analysis of the anomalous dry conditions that
characterised the campaign, the African easterly jet, easterly
waves and NWP data denial experiments.
Motivation
West Africa is a region that experiences marked interannual
variability of rainfall (e.g. Rowell et al, 1995). This impacts
water resources, agriculture and health, sometimes resulting in
extreme social and economic problems and loss of life. West African
climate variability also impacts the downstream tropical Atlantic
by influencing hurricane activity (e.g. Landsea and Gray, 1992).
Because of the marked variability and associated impacts there is a
clear need for skilful medium to seasonal range predictions of
rainfall and other variables for the West African region. There is
also a need for increased confidence in predicted climate change
scenarios for the region. The hydrological cycle in West Africa
shows a large sensitivity to projected climate change scenarios and
this at a time when water resources will come under intense
pressure from a fast-increasing population (IPCC, 2001).
GCMs are the major tools used for weather and climate
prediction. Current GCMs are hindered by large systematic errors in
the West African region. For example, GCMs are known to poorly
represent both the diurnal (e.g. Yang and Slingo, 2001) and annual
cycle of rainfall over West Africa (e.g. see West African Monsoon
Project final report (2001)). More process study work is required
to increase our knowledge and understanding of the processes that
determine the nature of West African weather and climate and its
variability. Where possible this needs to link with research aimed
at improving GCMs used for prediction. Focussed observing campaigns
are required to shed light on key components of the West African
monsoon that are not well observed by the routine network. One such
component is the African easterly jet (AEJ), a mid-tropospheric jet
present around 600mb and 15oN during the northern hemisphere
summer.
The AEJ, present across the whole of West Africa (fig. 1(a)),
has a key role to play regarding scale interactions in the West
African and Atlantic regions. Its vertical shear is known to
encourage the organised long-lived mesoscale convective systems
(MCSs) (e.g. Houze and Betts, 1981) responsible for most of the
daily rainfall events in the West African region. The
AEJ-associated potential vorticity gradients and low-level
temperature gradients satisfy the necessary instability criteria
thought to give rise to African easterly waves (fig. 1(b,c) (e.g.
Burpee, 1972), the major synoptic weather systems over West Africa
that also trigger Atlantic tropical cyclones downstream. Studies on
the interannual variability of West African rainfall have
identified coherent signals in the variability of the AEJ (e.g.
Newell and Kidson, 1984); whether the AEJ has an active or passive
role in such rainfall variability is unknown. Despite the
importance of the AEJ, it is poorly observed by the routine network
(c.f. Fig. 12)
The JET2000 project capitalised on the fact that in the summer
of 2000 the Met Office Research Flight (MRF) C-130 aircraft was
stationed in Cape Verde, and was available to make observations
over West Africa. Four flights with dropsondes, involving transects
along and across the jet and the baroclinic zone were made. These
observations are of unprecedented resolution for this part of the
world. The aim of this paper is to describe the motivation and
scientific background for the experiment and provide some
preliminary findings.
Scientific background
Recent work on convection over the tropical oceans has suggested
that boundary layer equivalent potential temperature (θe) and its
variations has a strong influence on the nature of convection and
circulations in the tropics (e.g. Emanuel et al, 1994). In this
context, the Hadley cell arises in association with large-scale
meridional gradients in boundary layer θe gradients, deep moist
convection preferentially occurring in the region of higher θe.
Zheng et al (1999) applied these ideas to the West African
monsoon. From their perspective, the West African monsoon arises in
association with marked north-south boundary layer θe gradients
that develop between the tropical land mass and the Gulf of Guinea
and South Atlantic ocean. The situation is complicated by marked
variations in surface properties over West Africa. North-south
variations in albedo and vegetation between the desert to the north
and rainforest to the south impact strongly on the surface fluxes
and the boundary layer θ and θe distributions (c.f. Nicholson et
al, 1998, Cook, 1999). As a result, peak values of boundary layer θ
and θe, which are largely collocated over the ocean, are instead
displaced over West Africa. θ peaks at around 25oN in the Sahara
(c.f. fig1(c)) while θe peaks further south in the rainy zone
around 10-15oN. Thorncroft and Blackburn (1999) discussed the
consequences of this on convection. Deep moist convection
characterises the high θe region while dry convection in the
Saharan heat low characterises the high θ region. They argue that
the mean tropospheric temperature profiles that are expected from
an adjustment to a moist adiabat in the deep moist convecting
region and a dry adiabat in the dry convecting region can help to
explain the observed decrease in meridional temperature gradient
with height, which in turn, through thermal wind balance explains
the observed midtropospheric AEJ.
Thorncroft and Blackburn's model makes simple assumptions about
the convection and the tropospheric temperature profiles. We need
to know more about the nature of these profiles from observations
and whether current GCMs can simulate or predict them. Observations
and modelling studies need to be used to investigate the
interaction between the dry and moist convecting regions. For
example, it is likely that mid-level dry intrusions from the Sahara
interact with the deep moist convecting zone affecting the nature
of the convection there, analogous to the dry intrusions identified
in the tropical West Pacific (e.g. Parsons et al 2000). Thorncroft
and Blackburn's model also represents a synthesis of a balanced
model and a turbulent convective model. Since the turbulence and
convection are controlled by the diurnal cycle in the boundary
layer, representation of the diurnal cycle seems to be crucial to
the larger-scale evolution, yet is known to be dealt with badly in
GCMs (e.g. Yang and Slingo, 2001).
If GCMs are to realistically predict the AEJ, they will need to
realistically represent the interactions between the land surface,
the boundary layer, the moist and dry convection and the dynamics.
We do not know how well this is achieved in current GCMs used for
weather and climate prediction because we have insufficient
observational data to describe these interactions. Specifically,
JET2000 was motivated by three science questions regarding the AEJ:
(i) What is the state of balance between the AEJ, the turbulent
convective boundary layer and the moist convecting atmosphere to
the south? (ii) How does the diurnal cycle of the convective
boundary layer influence the AEJ and monsoon flow? (iii) How do
variations in the land surface properties impact on the convective
boundary layer and the large scale dynamics?
While it is accepted that there is a strong, approximately
zonally symmetric, forcing of the AEJ in association with
meridional gradients in boundary layer θ and θe, it is well known
that the AEJ is unstable to growing synoptic scale AEWs. As
discussed in Smith et al (2001), very little research in recent
years has explored the morphology and dynamics of such tropical
weather systems that are of importance in daily to medium range
weather forecasting (apart from tropical cyclones). We agree with
Smith et al (2001) that forecasters in tropical regions have few
conceptual models and that there is a notable lack of useful theory
combining dynamics and moist convective processes. This is
especially true in the West African region where the situation is
hampered by lack of observational datasets. Operational forecasters
are forced to combine the outputs from deficient NWP models with
crude empirical rules. Our knowledge of the AEWs is dominated by
three sources: the GATE data (e.g. Reed et al. 1977); NWP analyses
(e.g. Reed et al 1988, Diedhiou et al, 1999); and theoretical
models (e.g. Thorncroft and Hoskins 1994ab, Paradis et al, 1995).
While Reed et al (1977) provided us with a valuable description of
AEWs, our view is dominated by the composite analyses close to the
West African coast. They provide a rather smooth picture, with
little information about the evolution of individual waves.
Operational analyses suffer from the sparsity of data going into
the models and the model systematic errors. Theoretical models have
helped us to develop conceptual models of AEWs but these must be
developed in conjunction with new observations.
Research is needed to improve the NWP models on synoptic space
and time scales, particularly in the representation of AEW
structure and convection. Alongside this, we need to develop better
conceptual models of AEWs for forecasting. Hence, in addition to
questions about the AEJ, JET2000 was motivated by three questions
about synoptic variability in the form of AEWs: (i) What is the
dynamic and thermodynamic structure of AEWs and how do they evolve
as they propagate along the AEJ? (ii) What is the relationship
between the AEW structure, thermodynamic instability and convective
rainfall? (iii) What is the impact of extra observations on NWP
analyses and forecasts of the AEJ, AEWs and downstream tropical
Atlantic?
The JET2000 questions are being addressed through a combination
of analysis of aircraft observations and modelling efforts
including data denial experiments. The MRF C-130 observations are
particularly valuable since they include high resolution boundary
layer measurements together with quasi-synoptic upper air
observations over scales of several thousand kilometres. Although
limited in temporal coverage to a period of less than one week, the
data obtained using this platform have provided a unique resource
with which to study the AEJ and AEWs. Some early results will now
be described together with some logistics for the experiment.
Logistics
The Met Office MRF C-130 aircraft makes high frequency wind,
thermal and humidity measurements, suitable for turbulence
diagnostics, as well as measuring various bands of upward and
downward radiation and observing certain aerosols and cloud
condensation nuclei. In addition to the aircraft observations in
JET2000, dropsondes were deployed for key sections of the flights,
at 0.5 - 1. degree resolution; in total 112 dropsondes were
deployed. Most were deployed successfully, although a minority of
sondes yielded degraded wind data, especially in the first flight.
During each flight the dropsonde data was transmitted to the data
centre at the Met Office in Bracknell. This enabled the Met Office
and ECMWF to assimilate the data in real time. The Met Office and
ECMWF have subsequently made analyses and forecasts without this
data to assess the sensitivity of including extra data in the
region.
The experiment was planned in parallel with two other operations
using the C-130; SHADE (Haywood et al, 2002) and SAFARI (Swap et
al, 1998). The aircraft was available for JET2000 in the period
25-31 August 2000, during which approximately 35 hours of flying,
on 4 flights, were available. This is the peak period for AEW
activity in the rainy zone (e.g. Thorncroft and Hodges, 2001) and
is close to the height of the Sahelian rainy season. Four flights
were made; two west-east flights between Sal and Niamey and two
principally north-south 'box' flights from Niamey (indicated in
fig. 4). Flight 1 (Sal to Niamey) and flight 4 (Niamey to Sal) were
intended to observe along-jet variability in association with AEWs.
They were flown at maximum altitude possible for the aircraft
(350-500mb), and deployed dropsondes at intervals of approximately
one degree of longitude. Flights 2 and 3 (the box) were mainly
aimed at making observations of the AEJ by deploying dropsondes
with a spacing of 0.5 - 1 degree of latitude, together with
aircraft measurements at low-levels. By flying twice over the same
box, assessment of AEW structure and of local temporal
variabilities was achieved. Low-level flights over the CATCH array
were included. Flight 3 reversed the course to flight 2, with the
low level flight preceding the high level box in order to estimate
changes over the diurnal cycle.
The dates of each of the flights together with the timings for
take off, landing and the times of the low-level flights are
provided in table 1. The number of sondes released is also included
for each flight. Flight 1 aimed to fly through an AEW trough before
it reached the ocean. Flights 2 and 3 aimed to observe different
phases of a wave while flight 4 aimed to fly through the same wave
giving us a 4-dimensional view of the AEW. Consequently flights 2,
3 and 4 were conducted on consecutive days.
Operational flight planning was initiated in September 1999.
Reconnaissance visits were undertaken to both of the planned
operating airfields – Niamey in Niger and Sal in Cape Verde. During
these visits, the planned flying was discussed with airport
authorities and Air Traffic Control Units, who were all supportive
of the project. There were to be a minimum of six and a possibility
of eight countries involved with over-flights (including the
release of dropsondes). The project was constrained in that
diplomatic clearances could only be officially applied-for in the
three weeks before departure. Given the amount of information and
different requirements needed by each country, this resulted in an
unusually large effort in exchanging informal information with the
appropriate British Embassies in the months prior to the
detachment. The local diplomatic representatives were able to
'sound out' their opposite numbers, in advance of the official
diplomatic requests. This process was assisted by strong support
from the local meteorological services. In the event, all of the
diplomatic clearances were officially granted, the final ones in
the week before the experiment!
The flight plans were reviewed on the day of the flights in
light of the meteorological situation for that day. Daily analyses
and forecasts were provided to the detachment by a dedicated
forecaster at the Met Office in Bracknell, as well as through
discussion with local forecasters. Up to date forecast information,
particularly in regard to the evolution of convection, was relayed
to the aircraft from Bracknell by SATCOM link.
Meteorological Situation During the Experiment
The JET2000 flights took place between 25th-30th August. Those
of us who were looking forward to experiencing our first Sahelian
squall line while in Niamey were disappointed. Between the time we
arrived in Niamey in the afternoon on the 25th to the time we left
in the morning on the 30th no rain fell on the city at all. In
fact, before JET2000 the last significant rain recorded at Niamey
airport was on the 17th August when 39.5mm fell. Although 2mm fell
on the 30th (after we had left), it was not until the 20th
September that more than 10mm fell again. Long term statistics for
Niamey airport (Le Barbe and Lebel, 1997) indicate a mean August
rainfall of 183mm and an average of 15 events (roughly one event
every two days). We were clearly in Niamey during a dry period.
In order to give a slightly larger scale perspective of this dry
period, we present in fig. 2 a time-series of the average rainfall
based on the 34 rain gauges in the EPSAT network (a roughly 1ox1o
square incorporating Niamey). The Niamey square was clearly
characterised by more frequent and intense rainfall events before
the start of the experiment. The last major storm of the season
occurred on the 17th August, when an average of 29.5mm fell. It is
also striking that for the period between the 24th August and 12th
September only one day was characterised by a rainfall event of
more than 5mm and most days had little or no rainfall at all
confirming the extremely long dry period. It turns out that this
dry period was felt over a wider region of the Sahel than just
southern Niger (see figs. 3 and 4). As reported in the October food
security bulletin of the USAID-funded Famine Early Warning System
Network, the transition from wet to dry conditions around the
middle of August lead to significant crop losses in the region
emphasising the vulnerability of societies in West Africa to
climate variability.
It is interesting to note that as well as the intraseasonal
variations in rainfall, the Sahelian rainy season of 2000 was also
characterised by intraseasonal variations in AEW activity. Although
this can be seen in the radiosonde time-series at Niamey airport
(not shown) it is most clearly illustrated in the time-longitude
hovmoller of band-passed filtered meridional wind at 700mb based on
ECMWF analyses (fig. 5). The Sahelian region experienced a sequence
of coherent west-to-east moving AEWs during the last half of July
to the middle of August, and then again in September. In between
these times, roughly coinciding with the dry period, the AEWs were
weaker and less coherent. This is consistent with the practical
difficulties that were encountered diagnosing the AEW phase during
the experiment. Whether this intraseasonal variability in
AEW-activity is consistent with the rainfall variability is an area
currently being investigated. Understanding this type of
intraseasonal variability in rainfall and AEW activity is important
for medium range weather prediction and may also be important for
understanding the West African climate variability on seasonal and
longer timescales.
We now present some preliminary analysis of the JET2000 period
including some of the observations made during the experiment.
The African Easterly Jet
The first high-resolution observations of the African easterly
jet structure were made on the second flight that took place on
28th August. Flight 2 consisted of a high level box between 8oN and
19oN (c.f. fig. 4 and fig. 12) with dropsondes released at
approximately whole degree intervals on the western side and half
degree intervals on the eastern side. This was followed by a
low-level flight at around 875hPa between 9.8oN and 16.5oN on the
eastern leg of the box. The AEJ observed on the eastern side, with
a peak value of -21.3ms-1 is clearly evident at around 675hPa and
10oN (figure 6(a)). The AEJ is much stronger than the
climatological average value of 15ms-1 often quoted and also
located about 5o of latitude equatorward of the expected latitude
for August (c.f. Reed et al (1977)). Newell and Kidson (1984) and
more recently Grist and Nicholson (2001) suggest that the AEJ is
stronger and equatorward of its mean position during dry Sahelian
years. Whether the strong and equatorward AEJ observed on flight 2
is consistent with the dry period in the region remains to be
determined.
Figure 6(a) also shows marked baroclinicity above and below the
AEJ consistent with the marked vertical shear there. The baroclinic
zone slopes downwards towards the Sahara and the strongest shear is
consistent with this. Above the sloping baroclinic zone, in
contrast, there is very little vertical shear in a region of low
static stability, consistent with mixing of momentum during dry
convection. The deep well-mixed layer above the surface at around
19oN is characterised by a potential temperature of about 315K.
Whereas at this latitude this well-mixed layer developed locally, a
well-mixed ‘wedge’ with similar potential temperature values exists
above the sloping baroclinic zone suggestive of equatorward
movement of air from the active dry convecting region towards the
moist convecting region. This is consistent with fig. 6(b) which
shows the humidity mixing ratio for the same section. Extremely dry
air characterises the low stability region around 19oN as expected;
but the section also indicates a sloping transition zone between
this dry air and the moister air equatorwards and beneath. A more
pronounced dry slot which may be the result of dry advection from
the Saharan region can also be seen sloping upwards from about
800hPa at about 14oN. More research is required, including
trajectory analyses, in order to identify more precisely the origin
of such dry slots; whether they originate from the Sahara or
midlatitudes or whether they arise from the actions of downdrafts
during moist convection. It is important to identify whether the
dry air seen in fig. 6(b) around 14oN originated from outside the
region since it may have had some role to play in the persistence
of the dry period in the Niger region.
Aircraft observations of the low-level θ and θe along this same
section made a few hours later are included in fig. 7 (as solid
lines). The low-level θ increases polewards as expected, with a
significant increase in the gradient polewards of about 12oN. The
low-level θe peaks around 12.5oN and generally decreases polewards
of this. As discussed above, these gradients are linked to the
north-south variations in the land surface, from a well-vegetated
surface with high soil moisture content in the south, to the Sahara
desert with sparse vegetation and low soil moisture in the north.
Associated with this, we expect north-south variations in
convection and associated vertical thermodynamic profiles. This is
illustrated in fig. 8, which shows tephigrams based on dropsondes
at 19oN and 8oN, latitudes we expect to be characterised by dry
convection and moist convection respectively. Consistent with this
we can see rather strikingly that the temperature profile at 19oN
is dominated by a deep dry adiabatic layer between 625mb and 850mb
with a new layer developing beneath. In contrast, the profile at
8oN is characterised by a profile closer to pseudoadiabatic with
higher humidities as expected. The result is that the two
temperature profiles cross at around 625hPa close to the observed
AEJ height in fig. 6(a) confirming, in a qualitative sense, the
conceptual model of Thorncroft and Blackburn (1999). However, the
profile at 8oN is warmer and drier than the saturated moist adiabat
with the local boundary layer value of θe proposed in this
conceptual model. The extent to which this typifies the region
during the season and the physics that determine these profiles
including the role of dry intrusions will be examined using GCM
analyses, available observations and idealised simulations.
An important goal of the JET2000 project is to assess whether
GCMs, used for weather and climate prediction, can adequately
reproduce the observed low-level gradients in θ and θe such as
those seen in fig. 7 and the atmosphere’s response to them such as
that illustrated in figs 6(a) and 8.
The Saharan Air Layer
A major feature of the West African monsoon, already illustrated
by the dropsonde at 19oN on the 28th August (fig. 8), is the deep
dry adiabatic layer that develops over the Sahara. This layer has
often been referred to as the Saharan air layer (SAL) (e.g.
Karyumpudi and Carlson, 1988) and has motivated research both due
to its impact on the West African monsoon but also its impact on
the Atlantic weather and climate (e.g. Prospero and Nees, 1986).
Indeed, the SAL was very clearly seen on the first flight out of
Sal on the 25th August. Leaving Sal, the aircraft profiled from
50ft above the ocean to approximately 500hPa. Figure 9 shows the
sounding for this aircraft ascent. It shows a very deep mostly dry
adiabatic layer above a cooler moist boundary layer consistent with
an overrunning of the oceanic boundary layer by the SAL. The deep
adiabatic layer was characterised by significant haze in
association with Saharan dust observed up to the inversion at about
520mb. It has been suggested that the dryness and high aerosol
content of the layer may be important for inhibiting moist
convection and tropical cyclogenesis in the tropical Atlantic
(Prospero, personal communication).
It is also important to note that the layer may also have a
dynamical impact. As discussed by Thorncroft and Blackburn (1999)
the low-static stability in the layer results in a very low value
of potential vorticity (PV). This can be seen in the August 2000
mean (fig. 1(b)) where low PV characterises the region polewards of
about 15oN over the continent but also over the tropical East
Atlantic. The low PV represents a significant negative PV anomaly
in the region and consistent with this, is associated with
anticyclonic relative vorticity on the poleward side of the AEJ.
Indeed, the low PV of the SAL is a key part of the PV-sign-reversal
that characterises the instability of the AEJ (e.g. Thorncroft and
Blackburn, 1999, Dickenson and Molinari, 2000).
The African Easterly Waves
Despite the weak AEWs during JET2000 (c.f. fig. 5) coherent
wavelike behaviour could still be diagnosed, especially towards the
end of the experiment. To illustrate these waves fig. 10(a) shows
the unfiltered meridional wind based on the ECMWF analysis at
700hPa on the 30th August. A sequence of AEWs is evident in this
analysis with well-defined troughs around 30oW and 12oW and a
broadscale trough around 5-10oE.
The trough over the ocean is well-defined and was associated
with the development of hurricane Ernesto a few days later. The
weak trough close to the West coast was observed during JET2000.
Flights 2 and 3 flew meridional transects in and behind it during
its passage past Niamey and flight 4 flew through it on the way
back to Sal. The trough east of this has a more complicated
structure with wind maxima at various latitudes between the equator
and 25oN. While one must be a little cautious of the analysis due
to the data sparsity of this region, the multi-centred nature of
AEWs is expected and is associated with AEJ-level PV anomalies and
low-level temperature anomalies (c.f. Pytharoulis and Thorncroft,
1999). The structure may be further complicated through the impact
of mesoscale convective systems (e.g. Houze and Betts, 1981) ,
orography (e.g. Mozer and Zehnder, 1996) or midlatitude trough
intrusions.
Given the relatively coherent analysis of the AEWs on the 30th
it is natural to wonder whether these waves were associated with a
coherent pattern of convection. The infrared satellite image at
this time (fig. 10(b)) indicates marked variations in convective
activity from west to east. The troughs diagnosed around 30oW and
12oW appear to have a marked convective signature. There is also a
broad region of convection east of this around 15oE but it is
uncertain whether this is linked to the passage of the broadscale
trough. More detailed analysis of the temporal evolution of AEWs is
needed in order to assess whether, in general, AEWs have a coherent
and predictable relationship with convection and to determine the
key processes that influence this relationship.
A Mesoscale Convective System
Planning for the third flight on 29th August was dominated by
discussion about an MCS approaching the southern sector of the box
from the east in the early morning. After discussions with the
forecasters at Niamey airport, it was estimated that the MCS would
intersect the box around 1100-1200 UTC. In order not to jeopardise
the high level box, it was decided to shorten the low level flight
to a transect from 14.1oN to 9.8oN, so that the aircraft could fly
north on the high level box while the MCS was in the vicinity of
the box to the south. Then the aircraft would close the box when
the MCS had moved to the west. This strategy also had the
attraction that we may be able to make observations both ahead of
and behind the MCS: an important consideration in assessing the
influence of the convection on the AEJ. In the event, the planning
was entirely successful: the MCS decayed over Benin, but
regenerated in the vicinity of the Atakora hills (just to the west
of the western leg, at around 11oN)) and observations were made in
advance of and behind the storm.
Figure 11 shows the north-south vertical section of zonal wind
and θ for the eastern side of the box which was made after the
passage of the MCS. There are significant differences between this
section and the same one on the previous day. The AEJ on this day
is still equatorward of what we may have expected at this time of
year but it has a different structure from that observed on flight
2. It has a split structure with peak values of 19.9ms-1 and
21.0ms-1 around 650hPa and 9.5oN and 12.5oN respectively. Between
these latitudes around 11oN there are marked easterlies around
825hPa and below which are consistent with the action of a
mesoscale downdraft Associated warming, as depicted by the bowing
down of isentropes, and drying (not shown) is further evidence for
the downdraft to the rear of the MCS. Figure 11 serves as a
reminder of the two-way interactions that can take place between
the AEJ and MCSs. While the vertical shear associated with the AEJ
is important for encouraging organised MCSs, the MCSs themselves
can also impact the AEJ. Whether GCMs can adequately resolve or
parametrise such interactions or whether they need to are open
questions that need investigating with parallel mesoscale and GCM
modelling studies.
Preliminary NWP Results
West Africa, along with most of the continent, is a well-known
data sparse region with very few routinely launched radiosondes.
Figure 12 shows the geographical location of all the sondes that
were assimilated into the ECMWF model at 12UTC on the 28th August.
The 25 dropsondes that were assimilated from flight 2 show up
rather dramatically. While the routinely launched radiosondes may
be useful for analysing along-jet variations (west of 10oE) at
around 14oN they are clearly inadequate for resolving the
significant north-south variations that exist in the region of the
AEJ. The significance of this for NWP over Africa and downstream
needs to be investigated: we aim to assess the impact of the extra
dropsonde observations made during JET2000 on analyses and
forecasts made using the operational ECMWF and Met Office
models.
Figure 13 shows the ECMWF model analyses with and without the
dropsondes assimilated. What is immediately striking is that the
ECMWF analysis without the dropsondes is actually quite good, the
AEJ maximum being within 24% of the strength observed and its
position being located to within 20 hPa and 0.9 degrees latitude
(c.f. Fig 6(a) and table 2). This is despite the fact that there
are no radiosondes in the vicinity of the observed AEJ. Including
the dropsondes in the ECMWF data assimilation strengthens the AEJ
by 1.6ms-1, shifts it to a new position now slightly equatorward of
the observed AEJ and creates a smoother-looking AEJ. The estimates
of vertical shear (in table 2) show quite good agreement between
both analyses and observations but this is due to compensating
errors with a weaker AEJ located slightly lower down than observed
and low-level easterlies that are too weak. The Met Office analyses
(summarised in table 2) also shows very good agreement with the
dropsonde observations. Including the dropsondes in the analysis
does not significantly improve the analysis of the AEJ-winds
because it was so close to the observations already (although other
aspects of the analysis may have improved, as in the ECMWF analysis
of low-level moisture indicated in fig.7 for example).
It would be wrong to speculate too much regarding the
differences seen here between the ECMWF and Met Office analyses
before more work is done. However it is clear that differences in
model sensitivities to the dropsonde data may arise from many
sources. These include model differences, assimilation differences
or differences in the total amount of data used (e.g. inclusion of
other data such as satellite observations) and random fluctuations.
Differences between the 3D and 4D variational assimilation schemes
used by the Met Office and ECMWF respectively were noted in the
analysis of FASTEX cases (Desroziers et al, 1999) and could also be
contributing to the differences seen here. The high quality of the
analyses without the dropsondes, despite the lack of routinely
reporting radiosondes (see fig. 12), may be fortuitous or may be
due to a good analysis of surface conditions and a realistic model
response to them. This is currently being investigated through more
data denial experiments.
Forecasts for 12UTC 28th August
The sparse routine observational network in West Africa means
that it is difficult to evaluate how well GCMs predict the AEJ and
associated AEWs. JET2000 observations give us an opportunity to do
this for the period of the experiment. We choose to illustrate here
a problem that both the ECMWF and Met Office models have with
predicting the AEJ several days in advance. The 5-day forecasts of
the AEJ in the JET2000 region for 12 UTC 28th August are shown in
fig. 14. In these forecasts, it is very difficult to see an AEJ at
all. Instead we see a region of weak broadscale easterlies with
erroneous mid-level easterlies north of 15oN. It is intriguing that
both the ECMWF forecast and the Met Office 5-day forecast have very
similar problems with a very weak AEJ and associated weak vertical
and horizontal shears. Assuming that the starting analysis for the
forecast (valid at 12UTC 23rd August) was as good as the analysis
for 12UTC on 28th August, then it is a concern that in the space of
5 days, both models can move so far from the observed and
climatological states. This clearly indicates that some processes
are being misrepresented in the region. Indeed, analysis of the
ECMWF 5-day forecast indicates that the low-level meridional
temperature gradient is too weak in association with an erroneously
cool and moist boundary layer polewards of about 13oN (see fig. 7).
This error is consistent with the systematic 5-day forecast errors
for August 2000 (not shown) emphasising the value and need for
process studies in this region.
In summary, both the ECMWF and Met Office models were able to
represent an AEJ which was considerably removed from its
climatological position, despite the absence of upper air
observations at this latitude. The study has also indicated
however, that the model forecasts appear to exhibit significant
drift away from model analyses and climatology. Ongoing work
associated with JET2000 aims to investigate these findings by
identifying those data streams which are enabling a good analysis
and by identifying the processes which explain the drift in the
model forecasts.
Final Comments and Future Plans
The observations from the JET2000 experiment are a valuable
scientific resource for study of this data sparse region of the
world and so all are available to scientific community without
delay (see BADC website at http://www.badc.rl.ac.uk/data/jet2000).
We still recognise the need for more observational studies in this
data sparse region. It is hoped that JET2000 research and
experiences gained will help in planning such experiments. Indeed,
a major international research and field campaign is currently
being planned for the West African region and will span several
years starting in 2004. This project has become known as AMMA
(African Monsoon Multidisciplinary Analysis) and is concerned with
weather and climate variability in the region and how this impacts
such things as water resources, food security, health, chemistry
and tropical cyclones. More details of the AMMA project can be
found at :
http://medias.obs-mip.fr:8000/amma/english/index_en.html
Acknowledgements
This experiment was funded by NERC under grant GR3/13118. C.
Thorncroft has been partially funded for this work by NSF (PTAEO:
1023911-1-24796).
We would like to acknowledge the support we received at the Cape
Verde and Niger Meteorological Services. In particular we would
like to personally thank Mr Soares from Cape Verde. We would also
like to thank the support and hospitality we received from ACMAD in
Niger, in particular Mr Boulaya and Mr Afiesimama.
We would also like to acknowledge the input of the aircrew and
aircraft scientists of the Meteorological Research Flight: Martin
Cook, Pat Coyle, Rob Gregory, Stuart Heath, Mike Kempster, Julian
Pantrey, Dave Pearce, Derek Percival, Martyn Pickering, Ken Quick,
Tony Simpson, Phil Summers, Ian Woodford and Paul Woodman. We would
like to acknowledge Jason Lander for computing support, Iain
Russell and TAMSAT for providing us with Meteosat imagery used
during the experiment and in fig. 10 and Richard Forbes from the
Met Office for supplying the code for the analysis of the
dropsondes. The OLR data used in figs. 3 and 4 was obtained from
UCAR.
References
Brovkin, V., Claussen, M., Petoukhov, V., and Ganopolski, A.
1998. On the stability of the atmosphere-vegetation system in the
Sahara/Sahel region. Journal Of Geophysical Research-Atmospheres
103(D24), 31613-31624 .
Burpee, R.W. 1972 The origin and structure of easterly waves in
the lower troposphere of North Africa. J. Atmos. Sci., 29,
77-90.
Cook, K.H. 1999 Generation of the African easterly jet and its
role in determining West African precipitation. J. Climate, 12,
1165-1184.
Desroziers, G., Pouponneau, B., Thepaut, J.-N., Janiskova, M.
and Veerse, F. 1999 Four-dimensional variational analyses of FASTEX
situations using special observations, Q.J.R.Meterol.Soc., 125,
3339-3358.
Dickenson, M. and Molinari, J. 2000 Climatology of sign
reversals of the meridional potential vorticity gradient over
Africa and Australia. Mon. Wea. Rev., 128, 3890-3900
Diehdiou, A., Janicot, S., Viltard, A., de Felice, P. and
Laurent, H. 1999 Easterly wave regimes and associated convection
over West Africa and the tropical Atlantic: Results from NCEP/NCAR
and ECMWF reanalyses. Climate Dynamics, 15, 795-822.
Dirmeyer PA and Shukla J. 1996. The effect on regional and
global climate of expansion of the worlds deserts.
Quart.J.Roy.Meteorol.Soc. 122: 451-482.
Emanuel, K.A., Neelin, J.D. and Bretherton, C.S., 1994 On
large-scale circulations in convecting atmospheres, Q. J. R.
Meteorol. Soc., 120, 1111-1143.
Goldenberg, S.B. and Shapiro, L.J. 1996 Physical mechanisms for
the association of El Nino and West African rainfall with Atlantic
major hurricane activity. J. Climate, 9, 1169-1187.
Grist, J. P. and Nicholson, S. E. 2001 A study of the dynamic
factors influencing rainfall variability in the West African Sahel,
J. Clim., 14, 1337-1359
Haywood, J., Francis, P., Osborne, S., Glew, M., Loeb, N.,
Highwood, E., Tanre, D. and Formenti, P. 2002 Radiative properties
and direct radiative effect of Saharan dust measured by the C-130
aircraft during SHADE: 1 Solar Spectrum. Submitted to JGR
Houze and Betts, 1981 Convection in GATE. Rev. Geophys. Space
Phys., 19, 541-576.
IPCC 2001: Climate Change 2001: Impacts, adaptation and
vulnerability – contribution of working group II to the third
assessment report of the Intergovernmental Panel on Climate Change
(IPCC), WMO, pp1000 (available at
http://www.grida.no/climate/ipcc_tar/wg2/index.htm).
Karyampudi, V.M. and Carlson, T.B. 1988 Analysis and numerical
simulations of the Saharan air layer and its effect on easterly
wave disturbances. J. Atmos. Sci., 45, 3102-3136
Landsea, C.W. and Gray, W.M. 1992 The strong association between
Western Sahel monsoon rainfall and intense Atlantic hurricanes. J
Clim., 5, 435-453.
LeBarbe, L. and Lebel, T. 1997 Rainfall Climatology of the
HAPEX-Sahel region during the years 1950-1990. J. Hydrol., 189,
43-73
Lebel, T. and LeBarbé L., 1997. Rainfall monitoring during
HAPEX-Sahel :2. Point and areal estimation at the event and
seasonal scales. Journal of Hydrology. 188-189, 97-122.
Mozer, J.B. and Zender, J.A. 1996 Lee vorticity production by
large-scale tropical mountain ranges Part II: A Mechanism for the
production of African Waves. J. Atmos. Sci., 53, 539-549
Newell, R.E. and Kidson, J.W. 1984 African mean wind changes
between Sahelina dry and dry periods. J. Climatol., 4, 27-33.
Nicholsen, S.E., Tucker, C.J. and Ba, M.B. 1998 Desertification,
drought, and surface vegetation: an example from the West African
Sahel. Bull. Amer. Met. Soc., 79, 815-829
Parsons,D.P., Yoneyama,K. and Redelsperger,J-L. 2000 The
evolution of the tropical western Pacific atmosphere-ocean system
following the arrival of a dry intrusion. Q.J.R.Meteorol.Soc., 126,
517-548.
Pedder, M. 1993 Interpolation and filtering of spatial
observations using successive corrections and Guassian filters.
Mon. Wea. Rev., 121, 2289-2902.
Prospero, J. M. and Nees, R. T. 1986 Impact of North African
drought and El Nino on mineral dust in the Barbados trade winds,
Nature, 320, 7345-738
Pytharoulis, I. And Thorncroft, C.D. 1999 The low-level
structure of African easterly waves in 1995. Mon. Wea. Rev., 127,
2266-2280
Reed, R.J., Norquist, D.C. and Recker, E.E. 1977 The structure
and properties of African wave disturbances as observed during
Phase III of GATE, Mon. Wea. Rev., 105, 317-333.
Reed,R.J., Klinker,E. and Hollingsworth, A. 1988 The structure
and characteristics of African easterly wave disturbances
determined from ECMWF operational analysis/forecast system.
Meteorol.Atmos.Phys.,38,22-33
Ringer, M.A., 1998 Tropical convective rainfall in the global
UM: Initial comparisons with estimates derived from Meteosat
infrared imagery, UK Meteorological Office, Forecasting Research
Technical Report No. 239.
Rowell, D.P., Folland,C.K., Maskell, K. and Ward, M.N. 1995
Variability of summer rainfall over tropical north Africa
(1906-92): Observations and modelling. Q.J.R.Meteorol.Soc., 121,
669-704.
Smith, R.K., Garden, G., Molinari, J. and Morton, B. (2001)
Proceedings on an International Workshop on the Dynamics and
Forecasting of Tropical Weather Systems. Bull. Amer. Met. Soc., 82,
2825-2829
Sperber,K.R. and Palmer,T.N. 1996 Interannual tropical rainfall
variability in GCM simulations associated with AMIP. J.Clim.,
9,2727-2749
Swap, B., J. Privette, M. King, D. Starr, T. Suttles, H.
Annegarn, M.Scholes and C.O. Justice (1998), SAFARI 2000: a
southern African regional science initiative,EOS Earth Observer,
10(6):25-28.
Taylor C.M. and Lebel T. 1998. Observational evidence of
persistent convective-scale rainfall patterns. Mon.Weath.Rev. 126:
1597-1607.
Thorncroft, C.D. and Blackburn, M., 1999 Maintenance of the
African easterly jet, Q. J. R. Meteorol. Soc., 125, 763-786.
Thorncroft, C.D. and Hodges, K.I. 2001 African easterly wave
variability and its relationship to Atlantic tropical cyclone
activity, J. Climate,14, 1166-1179
Thorncroft, C.D. and Hoskins, B.J., 1994a An idealised study of
African easterly waves Part 1: A linear view, Q. J. R. Meteorol.
Soc., 120, 953-982.
Thorncroft, C.D. and Hoskins, B.J. 1994b An idealised study of
African easterly waves Part 2: A non-linear view, Q. J. R.
Meteorol. Soc., 120, 983-982.
Yang, G. and Slingo, J. 2001 The diurnal cycle in the tropics.
Mon. Wea. Rev., 129, 784-801.
Zheng, X.,Eltahir, A.B. and Emanuel, K.A. 1999 A mechanism
relating tropical Atlantic spring sea surface temperature and west
African rainfall. Q.J.R.Meteorol.Soc., 125, 1129-1163.
Zipser, E.J. 1977 Mesoscale and convective downdraughts as
distinct components of squall line structure. Mon. Wea. Rev.,
1568-1589.
Figure Captions
Figure 1. Mean fields for August 2000 based on operational ECMWF
analyses: (a) Zonal wind at 700hPa with a contour interval of 2ms-1
and values greater than 6ms-1 shaded, (b) Ertel potential vorticity
at 315K (close to 700hPa at 15oN) with a contour interval of
0.05PVU and values greater than 0.2PVU shaded and (c) Potential
temperature at 925hPa with a contour interval of 2K and values
greater than 312K shaded. The positive and negative meridional PV
gradients associated with the PV maximum on the equatorward side of
the AEJ, combined with the positive meridional gradients in
low-level potential temperature satisfy the necessary condition for
mixed barotropic-baroclinic instability. See Thorncroft and Hoskins
(1994a) for more discussion of this and the significance for AEW
growth.
Figure 2 Time-series of the mean rainfall based on 34 raingauges
in the EPSAT square. The data is averaged over 24 hours and the
stations are located between 1.7oE and 3.1oE and 13.0oN and
13.9oN.
Figure 3. Time-latitude hovmoller showing outgoing long-wave
radiation (OLR) averaged between 10oW and 10oE. Shading indicates
OLR values less than 220Wm-2 with darkest shading indicating values
less than 140Wm-2. Low values of OLR are consistent with high
clouds which we use here as a proxy for rainfall. The figure
illustrates the weak convective activity over the Sahelian region
between the middle of August and the beginning of September.
Figure 4. A map of West Africa showing the difference between
the mean OLR (in Wm-2) for the period between 20th to 31st August
and a climatology of the same period (based on 1990 to 1999).
Positive differences over much of the region including the central
Sahel and Guinea Coast are indicative of relatively drier
conditions during the latter half of August and illustrate the
large area that experienced a dry period. Wetter conditions are
indicated in the West Sahel. Also included in the figure is a
schematic of the four flights made during the JET2000 experiment.
Flights 1 and 4 were between Sal and Niamey. Flights 2 and 3 were
‘box-flights’ starting and finishing in Niamey.
Figure 5: Time-longitude hovmoller showing band-passed (2-6
days) filtered meridional wind at 700hPa based on 00UTC and 12UTC
ECMWF operational analyses (includes dropsondes). For clarity only
southerlies greater than 2ms-1 are shaded. Dark shading indicates
values greater than 4ms-1. The figure illustrates the weaker
AEW-activity over West Africa between the middle of August and the
beginning of September.
Figure 6. Pressure-latitude sections of (a) zonal wind (shaded)
and potential temperature with a contour interval of 2K and (b)
humidity mixing ratio based on dropsonde data for flight 2 on 28th
August 2000. The section shown is for the eastern leg of the box
shown in fig. 2 along approximately 2.3oE. Location of dropsonde
release locations are indicated as red triangles. The objective
analysis used follows the successive correction method described by
Pedder (1993). The blackened regions along the top and bottom of
the figure denote the extent of the data coverage.
Figure 7. Meridional profiles of (a) potential temperature,θ and
(b) equivalent potential temperature, θe based on aircraft
observations on the 28th August at a height of approximately 875hPa
(solid), between 1317-1538UTC, ECMWF analysis for 12UTC 28th August
with the dropsondes (long dashed), ECMWF analysis for 12UTC 28th
August without dropsondes (small dashed) and the 5-day forecast
from ECMWF for 12UTC 28th August (dotted). The aircraft
observations were made along approximately 2.3oE approximately 4.5
hours after the observations presented in fig. 6 were made (c.f.
table 1), during which time the boundary layer was warmed by a few
degrees (consistent with offset). Data shown is averaged over 5
minutes. ECMWF data is based on a model level which varies between
890hPa (at 9N) and 860hPa (at around 17N) and is averaged between
1.5E and 3.5E.
The observed θ profile is characterised by a weak meridional
gradient up to about 12oN and a stronger positive gradient
polewards of this. The meridional θ gradient is generally well
captured by the analyses, perhaps consistent with the good analysis
of the AEJ. The meridional gradient in the 5-day forecast is weaker
in association with a cold error increasing with latitude polewards
of about 11oN.
The observed θe profile is characterised by a peak around 12.5oN
and a marked reduction polewards of this consistent with
increasingly drier conditions. Both analyses indicate significant
departures from observations polewards of about 14oN, with
anomalously high values and peaks around 16oN. The analysis without
the dropsondes is particularly poor with its anomalous peak even
higher than the observed equatorward peak. These errors are
associated with erroneously moist conditions at these
latitudes.
The meridional profiles of θ and θe in the 5-day forecast
exaggerate the errors seen in the analysis with large cold and
moist errors, especially polewards of about 13oN. The resulting
weaker θ gradient is consistent with the weaker predicted AEJ (c.f.
fig. 14).
Figure 8: Tephigrams showing temperature (solid) and dew point
temperature (dashed) based on dropsondes 19oN (bold) and 8oN
(normal) during the eastern leg of flight 2 on 28th August.
Figure 9. Tephigram showing temperature (solid), dewpoint
temperature (dashed) and winds for the aircraft ascending profile
from Sal on 25th August 2000.
Figure 10. (a) Operational ECMWF analysis at 12UTC 30th August
2000 of unfiltered 700hPa meridional wind with a contour interval
of 1ms-1 , (southerlies are shaded with dark shading for winds
greater than 4ms-1) and (b) the infrared satellite image at the
same time from METEOSAT.
Figure 11. Pressure-latitude section of zonal wind and potential
temperature based on dropsonde data for flight 3 on 29th August
2000. Zonal wind is shaded and potential temperature is contoured
with an interval of 2K. The section shown is for the eastern leg of
the box shown in fig. 4 along approximately 2.3oE. Location of
dropsonde release locations are indicated as red triangles. . The
objective analysis used follows the successive correction method
described by Pedder (1993). The blackened regions along the top and
bottom of the figure denote the extent of the data coverage.
Figure 12. Distribution of radiosondes (circles) and JET2000
dropsondes (triangles) in the north African region that were
assimilated into the ECMWF Model in the 12UTC analysis on 28th
August 2000.
Figure 13. Pressure-latitude sections of zonal wind averaged
between 1.3E and 3.3E for 12UTC 28th August 2000: (a) ECMWF
analysis with assimilated dropsondes, and (b) ECMWF analysis
without assimilated dropsondes. The model assimilation was
performed at T511 resolution, which is equivalent to a 0.3o grid
and had 60 levels in the vertical. The plot shown here uses a 0.5o
grid in the horizontal but includes the 60 levels.
Figure 14. Pressure-latitude sections of zonal wind averaged
between 1.3E and 3.3E for 12UTC 28th August 2000 and based on
T+120hour forecasts from (a) ECMWF and (b) The Met Office.
Table 1: Flight summary table, all times GMT.
Flight Code
Date
Take Off
Landing
Low-level
No. of sondes
1
25/8/00
09:09
15:25
None
21
2
28/8/00
07:01
16:20
13:17-15:38
35
3
29/8/00
06:58
15:03
07:14-08:43
34
4
30/8/00
09:55
15:45
None
22
Table 2: AEJ characteristics from observations and analyses.
Vertical shear is based on the zonal wind difference between the
peak easterly and the zonal wind at 925hPa.
AEJ
VARIABLE
DROPSONDES
RAW DATA
MET OFFICE
WITH
MET OFFICE
WITHOUT
ECMWF
WITH
ECMWF
WITHOUT
Maximum
-21.3ms-1
-19.1ms-1
-20.2ms-1
-17.8ms-1
-16.2ms-1
Height
650hPa
630hPa
630hPa
670hPa
670hPa
Latitude
10.0N
9.6N
9.8N
9.5N
10.9N
Vertical shear
Below AEJ
6.2 ms-1 /100hPa
5.3ms-1 /100hPa
6.0ms-1 /100hPa
6.4ms-1 /100hPa
6.4ms-1
/100hPa
�
http://www.atmos.albany.edu/facstaff/chris/wampreport_WORD.html
� � HYPERLINK http://www.lthe.hmg.inpg.fr/catch
��http://www.lthe.hmg.inpg.fr/catch�
� http://www.fews.net
� This peak value is taken from the raw data. The peaks in fig.
6(a) are weaker due to smoothing by the successive correction
analysis. See figure caption for more details.